presentation.knit

class: inverse center middle

# Reproducible Analytical Workflows in R

## Lindsey Dietz, PhD
### 2022 Iowa State Business Analytics Symposium
### April 5, 2022

### Download talk materials: github.com/iamaspacecoyote/2022_iowa_st_business_analytics

---

## Disclaimer

The views expressed in this presentation are strictly my own. They do not necessarily represent the position of the Federal Reserve Bank of Minneapolis or the Federal Reserve System.

---

## Who am I?

.left-column[
+ [Financial Economist/Stress Testing Production Function Lead @ Federal Reserve Bank of Minneapolis](https://www.minneapolisfed.org/news-releases/2021/dietz-recognized-for-excellence-in-bank-supervision)
    - I lead the team working on implementation, execution, and analysis of models for stress testing the nation's largest banks.

+ R Enthusiast and STEM diversity advocate
    - [Co-creator of the noRth R user conference](https://rnorthconference.github.io/)
    - [Co-organizer of the R Ladies - Twin Cities meetup](https://www.meetup.com/rladies-tc/members/)
    - [IF/THEN AAAS STEM Ambassador](https://www.si.edu/womensfutures)

+ Stathlete
    - [BS and MS in Math; Hall of Fame collegiate athlete at the University of Minnesota-Duluth](https://www.youtube.com/watch?reload=9&v=AzCXX9gH7ME)
    - PhD in Statistics from University of Minnesota-Twin Cities

]

.right-column[
<img src="data:image/png;base64,#img/statue.jpg" width="85%" />
]
---
class: inverse center middle

# Let's Dive In!

# https://github.com/iamaspacecoyote/2022_iowa_st_business_analytics

---
## Objectives of this discussion

- **Provide a starter pack on the R ecosystem to support report creation in HTML**

- Provide examples of table and graphics packages to display analysis

- Sprinkle in tips and tricks to make your reports dazzling and reproducible

---

## Self-study extensions (i.e. not covered)

- Setting up version control 
    - Please use it! Check out Jenny Bryan's [Happy Git and Github with R](https://happygitwithr.com/)

- Details of the tables or graphics packages 
    - These could easily be their own long form tutorials

---

## Can you relate to this process of reporting?

- Do analysis in one tool (Excel/R/Python/SAS/SPSS/Minitab/Stata/Matlab)

- Make tables and charts in another tool (Excel/Tableau/PowerBI)

- Screenshot or paste everything into a text editor (Word/LaTeX)

- Turn it into a final product (PDF/PowerPoint)
--
.center[
![Gif](data:image/png;base64,#https://media.giphy.com/media/2wZpm9zyceDyXHPf5S/giphy.gif?cid=ecf05e478g7s2h8a64fmbzty3bbgu6lh5ctbqmxh6t289p1t&rid=giphy.gif&ct=g)
]

---

## We can do it all with R!

---

## What is R?

- R is an **open source software** suite including
    - an effective **data handling** and storage facility,
    - a suite of operators for **calculations** on arrays, in particular matrices,
    - an integrated collection of intermediate tools for **data analysis and graphics**, and
    - a **programming language** which includes conditionals, loops, user-defined recursive functions and input and output facilities.

## R works best when combined with RStudio

RStudio is an **open source** integrated development environment (IDE) for R. It includes a console, syntax-highlighting editor that supports direct code execution, as well as tools for plotting, history, debugging and workspace management.

---

## [*Action Item*] Download and Install R & RStudio

- Install R 4.1.3 or higher
    - Navigate to a nearby CRAN mirror: https://cran.r-project.org/mirrors.html
        - Use the Iowa State mirror: https://mirror.las.iastate.edu/CRAN/
    - Choose the download for your operating system

- Install RStudio 2022.02.0+443 or higher
    - Navigate to https://rstudio.com/products/rstudio/download/
    - Choose the download for your operating system
---

## R packages

- The open source nature of R allows users to create software extensions called *packages*.
 - Packages are the fundamental units of reproducible R code and contain groups of useful functions.
     - [CRAN is a great place to start to find packages.](https://cran.r-project.org/web/packages/available_packages_by_name.html)
 - [You can make your own packages to support your work!](https://www.youtube.com/watch?v=J45Hs9XyN68)
 
 .center[
 ![](data:image/png;base64,#img/hex-stickers.png)
 ]

---

## [*Action Item*] Installing Packages

### [Install packages for this talk (Video)](https://youtu.be/_a9149G8LKs)
 - Open RStudio
 - Run the following command in your R console window

```r
install.packages(c('dplyr', 'tidyr', 'rmarkdown', 
                   'ggplot2', 'fredr', 'highcharter', 'plotly'))
```

---

## R users are an amazingly supportive community

.pull-left[
<img src="data:image/png;base64,#img/code_hero_rstats.png" width="100%" style="float:center" />
]
.pull-right[

- [R Ladies](https://rladies.org/) - worldwide organization to promote gender diversity
 
- R User Groups - in many major cities across the world

- R conferences - [rstudio::conf](https://www.rstudio.com/conference/), [useR!](https://user2022.r-project.org/), many local options
 
- [R Twitter (#rstats)](https://twitter.com/hashtag/rstats)
  - [Hadley Wickham](https://twitter.com/hadleywickham)
  - [Jenny Bryan](https://twitter.com/jennybryan)
  - [Dirk Eddelbuettel](https://twitter.com/eddelbuettel)
  - [Alison Presmanes Hill](https://twitter.com/apreshill/)
  - [Tom Mock](https://twitter.com/thomas_mock)
  - [Julia Silge](https://twitter.com/juliasilge)
]
---

## R <- Iowa State University
Two R visionaries did their PhDs in Statistics at Iowa State. 
.pull-left[
[**Hadley Wickham**](https://hadley.nz/), Chief Scientist at RStudio
Creator of the `tidyverse` (`ggplot2`, `dplyr`, ...)
![](data:image/png;base64,#img/hadley.jpeg)
]
.pull-right[
[**Yihui Xie**](https://yihui.org/), Software Engineer at RStudio
Creator of `rmarkdown` + workflow tools
![](data:image/png;base64,#img/yihui.jpg)
]

---
class: inverse center middle

# Making Dazzling Reports

---

![Artwork by @allison_horst](data:image/png;base64,#img/cracked_setwd.png)

---

## R Project

"R experts keep all the files associated with a project together — input data, R scripts, analytical results, figures. This is such a wise and common practice that RStudio has built-in support for this via projects."

-- Hadley Wickham ([R for Data Science](https://r4ds.had.co.nz/workflow-projects.html))

R projects define your working directories:

- Paths are relative rather than fixed
  - Your work is portable and tidy

---

## [*Action Item*] Open/Make an R project

### [Create a Project (Video)](https://youtu.be/CojhAWZ92FE)
 1. Open RStudio 
 2. Configure global options to avoid saving .Rdata (1-time step)
     - Tools -> Global Options -> Workspace -> Uncheck the box. Change the dropdown to Never.
     - Restart RStudio
 3. Create an R project to associate with a set of files

### [Open the Project for this talk (Video)](https://youtu.be/h6w5cn4xaKE)
 1. Download files from Github
 2. Click on R project file
 
---

## Folder Infrastructure

.pull-left[

- Store code (R scripts) separate from your reporting (R Markdown) files
 
- Consider pre-processing and storage of summary data 
 
- Don't try to fit too much into one report; separate reports can sit in one folder
 ]
 
.pull-right[ 
 
 ![Folder structure](data:image/png;base64,#img/folder.jpg)

]

---

.center[![Artwork by @allison_horst](data:image/png;base64,#img/rmarkdown_wizards.png)]

---

## What is R Markdown?

R code + Markdown (text-to-HTML conversion tool) = R Markdown

R Markdown files are designed to be used in three ways:

--
 - As an environment for you to do data science or analytics.
 
--

- Collaborating with colleagues with interest in your conclusions and how you reached them (the code).

--
 - **For communicating to decision makers, who want to focus on the conclusions, not the code behind the analysis.**

---

## [*Action Item*] R Markdown Walkthrough

- Open the dazzle.Rmd file inside the report folder
     - YAML Header & knitting with parameters
     - Code chunks
     - Markdown & LaTeX
     - File organization
     - Adding HTML and tables
     
 - [Download the R Markdown Cheatsheet](https://github.com/rstudio/cheatsheets/blob/main/rmarkdown.pdf)
---

## Dazzling Elements in R Markdown

- Using parameters in YAML header
 - Hiding code chunks and including 'Easter eggs' for other analysts
 - Using tabs and drop downs to improve pagination
 - Adding meaningful color to output
 - [Style output using HTML themes](https://www.datadreaming.org/post/r-markdown-theme-gallery/)

---

## Why HTML

.pull-left[
![](data:image/png;base64,#https://media.giphy.com/media/qDxe3pb4myxggnPe9u/giphy.gif?cid=ecf05e472wy9p3ozzzbrk4x3okwtsjygmurjaci8oorrmq7m&rid=giphy.gif&ct=g)
]

.pull-right[
- HTML is *interactive* without being Shiny
- HTML is *portable* and everyone you know has a web browser
- TeX overhead can be high for creation of pdf and RMD files may fail to knit in different environments 
]

---

## If you must PDF

- As long as you are not using LaTeX outside of R, you can use a minimal install of TeX via the `tinytex` package
```r 
install.packages('tinytex')
tinytex::install_tinytex()
```
 - [TinyTeX is a custom LaTeX distribution based on TeX Live that is small in size](https://yihui.org/tinytex/)

- YMMV: errors can accumulate especially if you don't have admin rights on your computer

---

## Building tabular content

There are many table packages. For HTML, `DT` and `reactable` are good starting points. For pdf, `kableExtra` is the gold standard.

.center[![](data:image/png;base64,#img/tidydata_7.jpg)]

---

## Building graphical content

`ggplot2` works great as a static graphics package. For interactive graphics, use packages like `plotly` or `highcharter`.

.center[![](data:image/png;base64,#img/presenting_monster.png)]

---

## Tips for Tables and Graphics

- Create tables and graphics to support a narrative
     + Storyboard your ideas on paper or a whiteboard before diving in
 - Build the tables and graphics outside of R Markdown first to avoid extra loading overhead
 - Think about your user's needs
     + Do they like to download the data?
     + Are you using color in an intentional way? (1 in 12 men are colorblind)

---

## Additional UX principles for reporting

Your report should:

- Have a clear hierarchy that makes navigation natural
 - Understand and embrace accessibility
 - Maintain time consistency - transition to or away from ideas slowly
 
---

## Extensions

- Version control of your reports
 - Deployment on a centralized server such as RStudio Connect opens up options such as scheduled reporting with email blasts
 - HTML reporting could be extended to true Shiny reporting if your users are asking for more interactivity

---

## References and Credits

- Artwork by Allison Horst sourced from [**Github**](https://github.com/allisonhorst/stats-illustrations)
 - [**R for Data Science**](https://r4ds.had.co.nz/workflow-projects.html) 
 - [**Advanced R**](https://adv-r.hadley.nz/)
 - [**Happy Git and Github with R**](https://happygitwithr.com/)
 - [**R Markdown Driven Development by Emily Riederer at rstudio::conf 2020**](https://www.rstudio.com/resources/rstudioconf-2020/rmarkdown-driven-development/)

Slides created via the R packages:

- [**xaringan**](https://github.com/yihui/xaringan)
 - [**xaringanthemer**](https://github.com/gadenbuie/xaringanthemer)

---
class: inverse center middle

# Thank you! Questions?